Apologies for creating a new post for this. I had tagged onto a previous post, may have got overlooked?

http://puredata.hurleur.com/sujet-6913-non-real-time-effects-processing

In any case the patch I attached in that post was a little incomplete.

Anyway I have created a granular upsample patch, which works well for time stretched audio and pitch shifted audio. This is based on the suggestion in the post linked above.
All works as expected apart from when I try to add in a reverb effect. In the example attached I have included one of the pd reverb examples, which I have modified by dividing the delay rates by 8. 8 being the fastest I can get the up sampling going in my patch without glitching. Unfortunately the results aren't as expected, the final output has more reverb than set, so my calculations are going wrong somewhere. I was wondering whether anyone had any suggestions?

Patch attached

http://www.pdpatchrepo.info/hurleur/upsamplegrain.zip

you've probably picked just about the HARDEST effect to try out with upsampling

i had a look, but it could be any number of things.

I'm not sure how the delaylines work when upsampled, but you are making very short delays of less than 1ms now, so that might cause weirdness?

I don't know why you scale the feedback by dividing by 2. Is that just arbitrary? If so, maybe try other values (like ?

my suggestion is maybe to just make a fresh copy of your patches in a new folder, and then set up a new test with a simple feedback delay instead of the reverb. If you can get that scaled and working correctly, it might give you a better idea how to fix the 'verb.

This may be beside the point, but I don't think it's really useful to oversample reverb.

Coasty is oversampling everything in his patch by 8 times to make some sort of faster than realtime file processor. I think in that case, the reverb also should be upsampled, to get the correct tone colour.

but yeah, it's hard.